File size: 1,759 Bytes
929da43 a986d1e 929da43 a986d1e 929da43 c7407f4 929da43 2477f18 929da43 2477f18 929da43 c7407f4 929da43 c7407f4 929da43 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 |
---
base_model: ad019el/Kabyle_xlsr-finetuned-tamasheq.en
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: Kabyle_xlsr-finetuned-tamasheq.en
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# Kabyle_xlsr-finetuned-tamasheq.en
This model is a fine-tuned version of [ad019el/Kabyle_xlsr-finetuned-tamasheq.en](https://huggingface.co/ad019el/Kabyle_xlsr-finetuned-tamasheq.en) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.4417
- Wer: 0.8444
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 3e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 200
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|:-------------:|:-----:|:----:|:---------------:|:------:|
| 6.087 | 3.41 | 300 | 2.8648 | 1.0 |
| 2.3631 | 6.82 | 600 | 1.3254 | 0.8705 |
| 1.0058 | 10.23 | 900 | 1.3736 | 0.8829 |
| 0.7859 | 13.64 | 1200 | 1.3322 | 0.8733 |
| 0.6834 | 17.05 | 1500 | 1.4417 | 0.8444 |
### Framework versions
- Transformers 4.33.0
- Pytorch 2.0.1+cu118
- Datasets 2.14.4
- Tokenizers 0.13.3
|